Variety Identification of Corn Seeds Based on Hyperspectral Imaging and Residual Mamba 1D CNN

Maize is a globally important crop, and reliable identification of seed varieties is vital for breeding and quality assurance. To overcome the limitations of conventional methods, this study developed a non-destructive approach integrating hyperspectral imaging (HSI) and deep learning. A Residual Mamba One-Dimensional Convolutional Neural Network (RM1DNet) is proposed, which integrates residual and Mamba modules to enhance feature learning. RM1DNet achieved 94.85% accuracy in classifying 20 maize seed varieties, outperforming traditional classifiers and baseline deep models. These results demonstrate the robustness and efficiency of RM1DNet, highlighting its potential to advance intelligent seed variety identification using hyperspectral data.